Premature Ventricular Contractions (PVC) are a cardiac arrhythmia that can be associated with an increased risk of adverse cardiac events such as ventricular arrhythmias and sudden death. Therefore, the characterization of this arrhythmia becomes crucial in the early diagnosis and prevention of possible life threatening cardiac diseases. In this paper a new approach for automatic detection of PVCs is presented, based on morphological derivatives and information theory techniques. Using these approaches a set of patient invariant features is introduced. Sensibility and specificity results (respectively, 96.35% and 99.15%) show the potential of the algorithm when applied to the MIT-BIH Arrhythmia database.